Burden of illness score for elderly persons: risk adjustment incorporating the cumulative impact of diseases, physiologic abnormalities, and functional impairments.

BACKGROUND/OBJECTIVES: To develop and validate a new risk adjustment index-the Burden of Illness Score for Elderly Persons (BISEP)-which integrates multiple domains, including diseases, physiologic abnormalities, and functional impairments. RESEARCH DESIGN SUBJECTS: The index was developed in a prospective cohort of 525 patients aged > or = 70 years from the medicine service of a university hospital. The index was validated in a cohort of 1246 patients aged > or = 65 years from 27 hospitals. The outcome was 1-year mortality. RESULTS: Five risk factors were selected from diagnosis, laboratory, and functional status axes: high-risk diagnoses, albumin < or = 3.5 mg/dL, creatinine >1.5 mg/dL, dementia, and walking impairment. The BISEP score (range 0-7) created four groups of increasing risk: group I (score 0-1), group II (2), group III (3), and group IV (> or = 4). In the development cohort, where overall mortality was 154/525 (29%), 1-year mortality rates increased significantly across each risk group, from 8% to 24%, 51%, and 74%, in groups I to IV respectively (chi(2) trend, = 0.001)--an overall 17-fold increased risk by hazard ratio. The c-statistic for the final model was 0.83. Corresponding rates in the validation cohort, where overall mortality was 488/1246 (39%), were 5%, 17%, 33%, and 61% in groups I to IV, respectively (chi(2) trend, = 0.001)-an overall 18-fold increased risk by hazard ratio. The c-statistic for the final model was 0.77. In each cohort, sequential addition of variables from different sources (eg, administrative, laboratory, and chart) substantially improved model fit and predictive accuracy. BISEP had significantly superior mortality prediction compared with five widely used measures. CONCLUSIONS: BISEP provides a useful new risk adjustment system for hospitalized older persons. Although index performance using different data sources has been evaluated, the full BISEP model, incorporating disease, laboratory, and functional impairment information, demonstrates the best performance.